Author: Tom Kerrigan
Date: 19:44:53 04/23/00
Go up one level in this thread
On April 23, 2000 at 22:28:59, Flemming Rodler wrote:
>On April 23, 2000 at 20:00:27, Tom Kerrigan wrote:
>
>>On April 23, 2000 at 19:35:06, John Coffey wrote:
>>
>>>I assume that at least some programs generate a complete move list at every
>>>branch of the tree. That is my intent at the moment. (My 1987 program was
>>>kind of dumb in that it only cared about move order at the base of the tree, so
>>>it always generated moves on the fly above the base without keeping a list.)
>>>
>>>So I see myself having a big array of moves. When I generate moves deeper than
>>>the base, I will add them to the array and keep track of where the beginning and
>>>end are for the current depth.
>>>
>>>Does anyone else do it this way? Is this a valid approach?
>>>
>>>John Coffey
>>
>>Yes, it's extremely valid. Here's what I do:
>>
>>move_t move_heap[];
>>int first_move[], last_move[];
>>
>>So to loop through all the moves for a given ply, this is what I do:
>>
>>for (i = first_move[ply]; i < last_move[ply]; ++i) {
>> // the move to examine is move_heap[i]
>>}
>>
>>-Tom
>
>Since you have that first_move[ply] = last_move[ply-1]+1 you don't need
>first_move[]. Maybe it will even save you some time since you do not avoid a
>lookup in first_move.
>
>/Flemming
Having first_move is nice because you can't access last_move[-1] (for ply 0). Of
course, you can avoid this problem if you're careful, but for now I'm taking the
straightforward approach.
-Tom
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.